You are on page 1of 23


2 Quantum states and observables 2.1 Quantum states of spin-1/2 particles . . . . . . . . . . . . . . 2.1.1 State vectors of spin degrees of freedom . . . . . . . . 2.1.2 Pauli operators . . . . . . . . . . . . . . . . . . . . . . 2.1.3 A preliminary interpretation of the boxes . . . . . . . 2.2 States and observables . . . . . . . . . . . . . . . . . . . . . . 2.2.1 Bras and kets . . . . . . . . . . . . . . . . . . . . . . . 2.2.2 Observables . . . . . . . . . . . . . . . . . . . . . . . . 2.2.3 Eigenvectors and eigenvalues . . . . . . . . . . . . . . 2.2.4 Position representation and wave functions . . . . . . 2.2.5 Momentum representation and Fourier transforms . 2.2.6 Combining quantum systems . . . . . . . . . . . . . . 2.3 Measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.1 Measurement postulate . . . . . . . . . . . . . . . . . 2.3.2 Expectation values . . . . . . . . . . . . . . . . . . . . 2.3.3 A proper explanation of the boxes . . . . . . . . . . . 2.3.4 Three readings of the Heisenberg uncertainty relation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 3 3 6 6 7 7 10 12 13 15 16 18 18 19 20 20




In order to do so.1. once we are done with this chapter. We will associate the situation of the spin of the particle pointing up (“spin up”) with the vector |0 . 3 .1 Quantum states of spin-1/2 particles State vectors of spin degrees of freedom Subsequently.1) whereas the spin of the particle pointing down (“spin down”) will correspond to |1 . its spin. and this is encoded in the 0 or 1. 2. we will abstract from the motion of the particles for a moment and merely speak about the internal degree of freedom of the particle. from this chapter on. At the risk of introducing a mild redundancy here. This notation is commonly used in physics nowadays. we will make use of the Dirac notation. now in a slightly more formal language. this chapter will contain a lot of material.Chapter 2 Quantum states and observables Let us start by discussing the Stern-Gerlach boxes of Chapter 1 again. we have pretty much covered all important structure elements of quantum mechanics. On the positive side. In other words. The reflect whether this spin is pointing up or down. On the negative side. (2. (2. we will primarily be concerned with applications – quite a consoling state of affairs. much of which may appear at first a bit alien and unusual. and then turn to a more general formalism. we will follow a didactical approach: We will first discuss the formalism at hand of our example of the above boxes.2) This is nothing to be afraid about: These are merely vectors.1 2. Alternatively.

1 |× = √ (|0 + i|1 ) 2 and | 1 = √ (|0 − i|1 ) 2 (2. respectively.5) (2. coming back to the above boxes. 1 |+ = √ (|0 + |1 ) 2 (2. We have said that |0 and |1 correspond to spin up and down. It is a superposition of the spin pointing up and down (omitting the hyphenation from now on). (2.12) . here with complex coefficients.4) but let us stick with the numbers for our purposes. More precisely. As we will see in more detail later. they are basis vectors in a complex vector space H. Here.7) (2. and Eq.6) C2 . They are called state vectors in quantum mechanics. We will see what this means very soon. they arise in a quite subtle fashion. The parameters α and β are hence complex. Or |ψ = sin(x)|0 + cos(x)|1 for x ∈ [0. and |↓ .10) are valid choices.7) does not correspond to the situation of the spin pointing either up or down. Similarly.9) (2. It is something very different. however: A situation of the form as in Eq. (2. For example 1 |+ = √ (|0 + |1 ) 2 is a legitimate choice. (2. QUANTUM STATES AND OBSERVABLES we could also have written |↑ . As such. not only vectors |0 and |1 make sense. but in fact any linear superposition |ψ = α|0 + β |1 ∈ H where α.4 CHAPTER 2.6) reflects normalization of the vector. (2. as we will see in a second. β ∈ C such that |α|2 + |β |2 = 1.3) (2. 2π ). ||ψ 2 = ψ |ψ = |α|2 + |β |2 .11) Superpositions we have already seen before. (2.8) (2.

as (|0 + i|1 )/ 2 and (|0 − i|1 )/ 2 are on the equator of the sphere.16) . As such. Again as a box: State vectors of spins: The state vector of a spin degree of freedom of a spin-1/2 particle can be written as a vector |ψ = α|0 + β |1 . if |ψ is a state vector and |φ is another state vector. (2. That is to say. |ψ and eiφ |ψ . (2. √ |0 and |1 correspond √ to the north and south √(|0 + |1 )/ 2 and (|0 + |1 / 2 as well √ pole of the sphere. This image is frequently employed in intuitive explanations. (2.2.14) Let us conclude this subsection by noting that since these vectors form a vector space. This representation is referred to as the Bloch sphere. (2.6) – up to their irrelevant global phase – can be identified with points in R3 on a unit sphere. satisfying |α|2 + |β |2 = 1. so in three spatial coordinates.13) with spin right.15) are identified with the same quantum states. These are all state vectors of a single spin degree of freedom. β ∈ C satisfying Eq. φ ∈ R. if we can say this now without sounding too confusing: this is a mild over-paramatrization of a state in quantum theory. But in fact. QUANTUM STATES OF SPIN-1/2 PARTICLES can be identified with spin left and 1 |− = √ (|0 − |1 ) 2 5 (2. Spin degrees of freedom are hence like “arrows” pointing into some direction.1. In fact. then any α|ψ + β |φ . any point on the sphere corresponds to a legitimate state vector. The global phase of a state vector and does not enter any prediction of quantum theory. we immediately come to a handy and intuitive picture representing a state vector of a single spin degree of freedom: Numbers α.

the system will be in an eigenvector of the respective observable.19) We hence note that the vectors |0 and |1 are eigenvectors of σz : Up to a complex number – here +1 and −1. see below.20) (2. Pauli operators are examples of Hermitian operators. the respective eigenvalues – we obtain again the same vector if we apply σz to it. in fact with Hermitian operators. (2. For example. Similarly. The short and somewhat elliptic explanation of the above situation involving the boxes is as follows: Properties are associated with operators.23) 2. A measurement along the z axis corresponds to the Pauli-z -matrix. let us consider an example: The Pauli-z -matrix σz acts as σz |0 σz |1 = = |0 . (2. So the first measurement corresponds to a “measurement of the observable Pauli-z ”. Before we become too abstract at this point. and similarly for the other Pauli matrices. QUANTUM STATES AND OBSERVABLES with α. = −|− .2 Pauli operators We have seen – even if not discussed in all detail – that the internal states can be associated with vectors. The above mentioned state vectors that are on the north and south poles of the Bloch sphere hence correspond to eigenvectors of σz with the respective eigenvalues. (2. the outcome of the measurement being the eigenvalue of the eigenvector. we find that the Pauli-x matrix σx acts as σx |+ σx |− = |+ . −|1 . (2.1.24) . 2.6 CHAPTER 2. the Pauli-y -matrix σy has the property that σy |× σy | = = |× .17) where again A|ψ ∈ H. we measure the spin along the z axis. If we get the value +1. hence “measure the observable Pauli-z ”. we will obtain the state vector |0 = σz |0 (2. Such Hermitian operators are called observables.1.18) (2. is again a legitimate state vector. so with linear maps of the form |ψ → A|ψ . appropriately normalized. −| .21) Finally.22) (2. β ∈ C.3 A preliminary interpretation of the boxes We will explain this in more detail below. We will now see that physics properties can be associated with operators A. After the measurement.

we can surely guess how a general state vector of such a quantum system would look like.27) (2. 2. Neither |0 not |1 are eigenvectors of σx .2. we obtain the state vector |1 = −σz |1 (2. the situation is quite different.26) in case of the other. We will see that this fact is essentially responsible for the two outcomes. we had two basis vectors |0 and |1 . After this teaser. i. After the measurement.2.2. In fact. Since |0 is an eigenvector of σz .e. for example. In case of the value −1. if we will repeat measuring σz . For example. Let us hence say that we have d levels.25) after the measurement. usually. If we at some point measure along the x axis. there are situations in physics where one has a larger number of basis vectors. which could be counted in one way or the other. This is the most general form of a superposition. atoms have a large number of energy levels. But having said that. STATES AND OBSERVABLES 7 after the measurement. being obtained in a probabilistic fashion.. corresponding to spin up. in contrast to so-called infinitedimensional quantum systems (the position of a particle does. where d is any natural number. This is also called an finite-dimensional quantum system. This is why then the spin is always pointing into the same direction. we are in the position to consider this situation a bit more systematically and more carefully. . the two levels could not only represent the spin degree of freedom. the state vector is an eigenstate of the observable measured. This could be the two energy levels of an atom. At this point.1 States and observables Bras and kets In the above example. but in fact any two internal degrees of freedom.2 2. Needless to say. we will repeatedly get the outcome +1 and the post measurement state vector |0 . spin left and spin right. 1 1 |+ = √ (|0 + |1 ) = σx √ (|0 + |1 ) 2 2 in case of one outcome and 1 1 |− = √ (|0 − |1 ) = −σx √ (|0 − |1 ) 2 2 (2. there is no need for this number necessarily being two. does not correspond to a finite-dimensional quantum system).

|d − 1 } span the Hilbert space H Cd . Very important will also be the scalar product. Hilbert space of d-dimensional quantum systems: The basis vectors {|0 . . (2. in fact a Hilbert space (see Appendix). |d − 1 } (2. . j =0 (2. .31) we write their standard scalar product as d−1 ψ |φ = φ|ψ ∗ = j =0 ∗ αj βj . Scalar product: For two state vectors d−1 d−1 |ψ = j =0 αj |j . . The principle that any of these vectors are allowed state vectors is sometimes also referred to as the superposition principle.30) are again orthonormal basis vectors.8 CHAPTER 2. QUANTUM STATES AND OBSERVABLES State vectors of finite-dimensional quantum systems: The state vector of a d-dimensional quantum system can be written as a vector d−1 |ψ = j =0 αj |j . while all other state vectors are superpositions thereof. Dual vectors. These vectors form a vector space H. . .28) satisfying normalization d−1 |αj |2 = 1. the vectors {|0 . . are referred to as “bras”. (2.29) Indeed. (2. judge for yourself – are called “kets”. d−1 ψ| = j =0 βj j | .33) . |φ = j =0 βj |j (2. which is why vectors – in an instance of physics humor. . from the dual space H∗ .32) Such scalar products are sometimes called “brackets”.

but by the ray associated with this state vector: Rays as quantum states: The ray of any state vector |ψ ∈ H is defined as {λ|ψ : λ ∈ C}.34) As always.40) . . Again.37) j =0 This may at this point look a bit awkward. so the state of a quantum system is strictly speaking not defined by the state vector. this fact is often silently omitted. . in order not to make the notation overly bombastic. (2. (2. we have a d-dimensional complex vector space with an orthonormal basis.38) Once one has grasped this.2. d − 1. j =0 (2.39) with    α0 . They are complete. (2.35) By construction. and equal to unity in case of a normalized state vector.36) for j. We have that the basis vectors are orthogonal. j |k = δj. STATES AND OBSERVABLES and again they are normalized if d−1 9 |βj |2 = 1. in that the span of these vectors is the entire Hilbert space. Summarizing this subsection. d−1 |j j | = 1 .k (2. . So we can associate d−1 |ψ = j =0 α j |j (2. this is always a non-negative real number. (2. αd−1   . . the global phase does not enter any prediction. . but we will see how useful this completeness relation is in a minute. which is a bit imprecise in that the global phase is not defined in a state. . k = 0.2. This fact one usually speaks of state vectors as of “states”. the scalar product induces a norm of the vector d−1 |ψ 2 = ψ |ψ = j =0 |αj |2 . or equivalently.

A = A† .k = j |A|k . we formulate the matrix form of the Pauli operators. |φ .44) for all |ψ . so “observable quantities”. QUANTUM STATES AND OBSERVABLES d−1 ψ| = j =0 βj j | (2. Such observables can again be written in matrix form: In terms of the basis {|0 . βd−1 ] . (2. .42) 2.2 Observables Observables. .47) The vectors |0 and |1 are already eigenvectors of σz . are Hermitian operators.k j |A|k |j k |. so we should not be surprised to see that the matrix form is diagonal. (2. (2.46) In the following. The first one is very much obvious: We can write σz = 1 0 0 −1 .41) with [β0 . . . This is such an important statement that we give it a box: Observables: Observables in quantum mechanics correspond to Hermitian operators. . As an exercise. (2. This is so common and natural that any other choice would unnecessarily make the notation very cumbersome. we will always identify an operator with its matrix form. entities that capture a measurement prescription. This is a very common identification: We will write A both for the linear operator itself as well as for the matrix that reflects it given a basis. (2. |d − 1 }. the matrix form has the entries Aj.45) It is clear that this matrix form uniquely characterizes the operator: We can make use of the “insertion trick” A=1·A·1= j. .10 and a dual vector CHAPTER 2.2.43) The adjoint can be defined via ψ |(A|φ ) = ( ψ |A)|φ (2. we find the matrix . Similarly.

that no two observables necessarily commute.51) The order of multiplication matters! To make this difference manifest. We collect them in a box: Pauli matrices: σx = 0 1 1 0 .48) We have included in this list the identity 1 to the matrix form of the Pauli operators.2. This basic feature of the formalism is at the root of the observation that orders of measurements matter. a rather formal observation: The Pauli matrices form a basis of all Hermitian 2 × 2 matrices: In other words. of course. We note a number of things: First. .2. σz ] = σx σz − σz σx . Commutator: For operators A and B . (2. σz = 1 0 0 −1 . β. every Hermitian 2 × 2 matrix A can be written as A = ασx + βσy + γσz + δ 1.49) with real α. 11 1= 1 0 0 1 . γ. one considers the commutator [σx . δ . (2. This definition we emphasize here simply because this notion is so often used in the literature.53) Properties in the quantum world correspond to operators that do not commute. the Pauli matrices. It is also the key to the understanding that it does not make sense in quantum theory to think of two different quantities corresponding to non-commuting observables to “take specific values”. This does not mean.50) . They do not. B ] = AB − BA. primarily for reasons of convenience. σy = 0 i −i 0 . (2. (2. A profound insight is that these matrices do not commute: It is by no means the same whether we compute σx σz = or σz σx = 0 1 0 −1 −1 0 1 0 (2. the commutator is defined as [A. STATES AND OBSERVABLES form of the other two Pauli operators.52) which is in general different from zero. (2.

(2.54) 2. Eigenvalue decomposition of observables: Every (finite-dimensional) observable A can be written as d−1 A= j =0 λj |φj φj |. we state at this point an equivalent form of the eigenvalue decomposition of observables: Diagonalization of observables: Every (finite-dimensional) observable A can be written as A = U DU † . and U is unitary. . Since eigenvalue decompositions are so important. The above statement is hence a manifestation of the observation that in their eigenbasis. . . they are called degenerate. which is no longer one-dimensional. . . λd−1 ∈ R and the eigenvectors {|φj } can be chosen to form an orthonormal basis of H. The respective eigenvalues then span the eigenspace. . This does not mean.12 CHAPTER 2. observables are diagonal. so satisfied U U † = U † U = 1.56) where D = diag(λ0 . (2. the fact that observables are Hermitian means that their eigenvalues are real. λ0 . . that all of the {λj } are necessarily different. . B ] = 0. .2. and correspond to basis changes of orthonormal basis. The . QUANTUM STATES AND OBSERVABLES Compatible observables: Two observables A and B are compatible. If two or more eigenvalues are identical. if [A. (2. Eigenvalues and eigenvectors play a very important role in quantum mechanics: The former essentially as “measurement outcomes” and the latter as “postmeasurement state vectors”.55) where the eigenvalues satisfy λ0 . (2. λd−1 ). . of course. λd−1 ∈ R.3 Eigenvectors and eigenvalues Indeed. . .57) Unitary operators are those that preserve scalar products.

Let us now do the opposite and let us look at a quantum mechanical description of a spatial degree of freedom alone. . without saying anything wrong at any instance in time. z . (2. This is a helpful insight. STATES AND OBSERVABLES above notation is also common for d−1 13 D = diag(λ0 . (2.4 Position representation and wave functions Before we finally come to the role of measurement and can carefully and properly explain our above setting involving boxes. . This may be taken as an intuitive approach. (2. Let us start with the operators reflecting position and momentum: Position and momentum: Position and momentum are reflected by the Hermitian position and momentum operators X and P . is probably already quite familiar. We have neglected the spatial degree of freedom and have looked at the spin degree of freedom alone.2. . Also. λd−1 ) = j =0 λj |j j |. (2. and hence all eigenvalues are ±1. This is even true for slightly more general operators: The same statement in Eq. The eigenvalues are hence distributed on the unit circle in the complex plane. tr(σi ) = 0 for i = x.58) Hermitian observables can hence always be unitarily diagonalized. . and will try to steer away from any fine print. .56) is true for so-called normal operators A which satisfy AA† = A† A. Most quantum mechanics books start here. We also emphasize an important insight: Observables can be simultaneously (unitarily) diagonalized if and only if they commute. y. in particular when one thinks of the diagonalization of unitary operators U (which are in general not Hermitian): This can be written as d−1 U= j =0 eiφj |j j |.2. allowing for an interpretation of a density.2. as ideas of particles flying around in free space is very intuitive. Unfortunately. the idea of a wave function. the treatment of position and momentum is somewhat overburdened with mathematical technicalities.60) 2. We take a pragmatic approach here. if one wants to do things rigorously.59) with all φj ∈ R. We end this subsection with a note again on the (three. different from 1) Pauli matrices: Their determinant and their trace are given by det(σi ) = −1. let us turn to another important aspect of the above situation.

14 CHAPTER 2. is again a state vector. We can at this point already grasp why this is the case: eigenvalues correspond to measurement outcomes. these operators no longer have discrete spectra. (2. Again.61) even if at this point we take the right hand side to be the definition for the left hand side of the equality sign (we do not insist the dual vectors x| to be contained in any dual Hilbert space. as it also contains phase information. both the position and the momentum operator have R as their spectrum. the Hilbert space is H = L2 (R) of square integrable functions over the reals.62) They satisfy ∞ |ψ (x)|2 = 1. It still makes sense for x ∈ R to write x|ψ = ψ (x).64) for x ∈ R has a probability interpretation. is that the probability density of finding particles in particular places can be determined from the wave function. a particle can take a continuum of different positions. ρ(x) = |ψ (x)|2 (2. the wave function is “a lot more than a mere probability distribution”. this is the entire real line. we will come back to that). if ψ1 : R → C and ψ2 : R → C are wave . we merely do not know where it is”. In case of position. QUANTUM STATES AND OBSERVABLES Now. however. then any linear combination of them. And in fact. Neither the position nor the momentum operator have eigenvectors (although many books spell them out nevertheless. in case of a onedimensional problem. This statement – even if one can sometimes read such statements in the literature – is wrong. state vectors are elements of Hilbert (and hence vector) spaces: That is to say that if |ψ1 is a legitimate state vector and |ψ2 as well. −∞ (2. In other words. This representation in terms of wave functions is usually called the position representation. Wave functions: Single spatial degrees of freedom are described by wave functions ψ : R → C. the wave function does not reflect the situation that “the particle is somewhere. as it is a complex-valued function. Because we ask for ψ being square integrable. so no discrete eigenvalues. Stated in terms of the position representation. So indeed. (2. Yet. What is true. appropriately normalized.63) |ψ (x)|2 can be interpreted as the probability density for finding the particle at position x. but we start worrying about that detail in the appendix). as we now have dim(H) = ∞.

One has H = span{|0 .2. suitable prepared. (2. will show the interference pattern. . This is often – and famously – written as [X. }.70) Some caution is needed here. as neither the position nor the momentum operators are bounded operators. |1 .69) where the bar marks that the Hilbert space is given by the closure of the span of the basis vectors. . so the function with values 1 ψ (x) = √ (ψ1 (x) + ψ2 (x)) . but this is no longer finite-dimensional. .67) N This is different from 1 √ |ψ1 (x)|2 + |ψ2 (x)|2 .65) N= −∞ dx |ψ1 (x) + ψ2 (x)| . 2 (2. STATES AND OBSERVABLES 15 functions. . Again. 2. one can take an orthonormal basis of the Hilbert space H of the spatial degree of freedom of a particle. A single particle. we will offer a fix for that in the appendix. (2. Note also that this is not a statistical interference because of many particles coming together or so.66) This is by no means a detail: The probability density ρ(x) of finding a particle at x is now given by ρ(x) = = 1 √ |ψ1 (x) + ψ2 (x)|2 N 1 ∗ ∗ √ |ψ1 (x)|2 + |φ2 (x)|2 + ψ1 (x)ψ2 (x) + ψ1 (x)ψ2 (x) .2. (2.5 Momentum representation and Fourier transforms Position and momentum do not commute. but infinite-dimensional. N (2.68) in that new terms arise reflecting the superposition. P ] = i . ∞ (2. Again. new features emerge. N where N reflects normalization. so is ψ = ψ1 + ψ2 (up to normalization).2. This is exactly the feature that is seen in the familiar double slit experiment: One does not merely see the joint pattern of each of the preparations in a superposition: But in fact. reflecting the fact that the complex phases matter.

|d2 − 1 }. we can introduce the momentum representation: Similarly as above. .77) (2. . Or we aim at describing several different particles at once. |d1 − 1 }. .76) (2. .75) This looks more complicated than it is: While an arbitrary superposition of a state vector from H1 can be written as d1 |ψ1 = j =0 αj |j (2. The Hilbert space of the joint system is then given by the tensor product H = H1 ⊗ H2 . QUANTUM STATES AND OBSERVABLES Similarly to the position representation. d1 − 1. .16 CHAPTER 2. .79) . second degree of freedom. .72) 2. . where ψ : R → C is the Fourier transform of ψ 1 ψ (x) = √ 2π ∞ (2. It is spanned by the orthonormal basis vectors {|j ⊗ |k : j = 0. (2. . . d1 − 1. we consider ˜(p). (2. . Such basis elements of tensor products are sometimes also written as {|j.78) and an arbitrary superposition of a state vector from H2 is d2 |ψ2 = j =0 βj |j . Let us assume that we have one degree of freedom associated with a d1 -dimensional Hilbert space H1 = span{|0 . . . capture all superpositions of two spin degrees of freedom of two particles described by quantum mechanics. . . . coming along with a d2 dimensional Hilbert space H2 = span{|0 . (2.6 Combining quantum systems How do we describe composite quantum systems in quantum theory? Clearly. k : j = 0. . . . the formalism must have an answer to that. . p|ψ = ψ ˜ : R → C. d2 − 1} . d2 − 1} . ψ −∞ (2. . k = 0.74) These spaces could. . We think of a particle having several degrees of freedom. for example. k = 0. (2.73) We then consider another. How do we capture this situation? Composition of degrees of freedom is incorporated by the tensor products in quantum mechanics.71) ˜(k )eikx dk. .2. .

it is. needless to say. on the first page of this script we have already encountered a composite quantum system. with all γj.k cj. even if at this point we had abstracted from this fact. for more elaborate composite quantum systems having many parts. an arbitrary linear operator can be decomposed as O= j.83) This is not such an alien situation in quantum physics: We should not forget that most things we can see are quantum systems with more than 1023 individual parts. If you think at this point that it may be confusing that such general state vectors contain ones that are no longer a product between the respective Hilbert spaces: Indeed. STATES AND OBSERVABLES 17 an arbitrary state vector taken from the composite Hilbert space H = H1 ⊗ H2 is given by d1 d2 |ψ = j =0 k=0 γj.k ∈ C.k |j ⊗ |k . (2.2. respectively. In fact.81) Similarly. and we will come to the profound implications of this later.80) as a linear combination of all new basis vectors. (2. Again: Composite quantum systems: The Hilbert space of the composite quantum systems the parts being associated with Hilbert spaces H1 and H2 is given by the tensor product H = H1 ⊗ H 2 .k Aj ⊗ Bk .84) The Hilbert space of the three coordinates of the spatial degree of freedom of a single particle is given by H = L2 (R) ⊗ L2 (R) ⊗ L2 (R). (2. (2. The parts of composite quantum systems also do not have to be the same. So the joint Hilbert space capturing this situation is given by H = L2 (R) ⊗ C2 .2. the Hilbert space of n particles of the same character (and associated with H each) is given by H ⊗ · · · ⊗ H = H⊗n .82) with operators {Aj } and {Bk } on H1 and B1 . (2.85) . (2. The same is true. For example. The particle has a position and a spin degree of freedom.

3 Measurement After all this. The situation of encountering degenerate eigenvalues merely requires a mild modification: In case of a degenerate observable A. Then the probability pj of obtaining the outcome λj is given by pj = | ψ |φj |2 .90) . QUANTUM STATES AND OBSERVABLES 2. 2.1 Measurement postulate Let us start by precisely stating how measurement is captured in quantum theory.3. . we are finally in the position to carefully and precisely describe the boxes that we have encountered above. . . the measurement values are the eigenvalues. . . . and a measurement “collapses” the state vector into one of the eigenvectors. we will have a look at a more general framework to capture measurement. . with eigenvalues λ0 . .89) where πk = j.λj =µk | φj φ j | (2.86) are degenerate. pj (2. For doing so.87) the state vector immediately after the measurement is given by 1 |ψj = √ φj |ψ |φj . we need to precisely understand what measurement does to a quantum system. µD−1 . and then discuss what this means in great detail.88) In other words. (2. Let us assume that none of the eigenvalues λ0 . λd−1 and different eigenvalues µ0 . . λd−1 of d−1 A= j =0 λj |φj φj | (2. which is however not fundamentally more general than the measurement postulate that we are going to see now).18 CHAPTER 2. Measurement postulate (for von-Neumann measurements): We consider a measurement of the observable A for a quantum system prepared in a state vector |ψ ∈ Cd . . . we can write D −1 A= k=0 µk πk (2. The subsequent type of measurement is called “von-Neumann measurement” (to emphasize that later on. .

is given by p k = ψ |πk |ψ . this prescription simply becomes the previous one.3. we have +|σz |+ = 0. As these are projections.97) (2. we have πk πk 2 = πk .93) the state vector immediately after the measurement is given by 1 |ψk = √ πk |ψ .2 Expectation values Needless to say. 2. if we measure a Pauli operator σz on a system initially prepared in |0 . pk (2. k = 0. The expectation value is of key importance in quantum mechanics: is the expected value of the averaged outcomes of a single type of measurement. Then we will obtain the expectation value 0|σz |0 = 1.2. (2. . In case of |1 . And in case of |+ as defined in Eq.7). Expectation values: The expectation value of an observable A for a state vector |ψ is given by A = ψ |A|ψ . .94) Clearly. (2.98) (2. .92) = † πk for all k .96) Again.3. Then the probability of obtaining the outcome µk . for non-degenerate observables.95) Because observables are Hermitian. expectation values are always real. . D − 1. MEASUREMENT 19 is for each k the orthogonal projection onto the subspace spanned by the eigenvalues that take the value µk . a single run of the experiment will not provide us the expectation value. (2. a single measurement will not provide us with complete information about the probabilities {pj } – similarly to a single coin being tossed will not say whether the coin is fair. (2.91) (2. but the averaged outcomes of many runs will asymptotically converge to this value. For example. (2. we get 1|σz |1 = −1. .

and we can repeat the measurement and arbitrary number of times.101) 2 So the probability of getting each outcome is precisely 1/2. There are few statements in quantum theory. (2. to which we will come later. Having understood the measurement postulate. Then the probability of getting the post-measurement-state |+ is given by | +|0 |2 = and in the same way 1 .20 CHAPTER 2. 2 (2.3 A proper explanation of the boxes We are now in the position to properly explain our initial situation involving boxes – with the single exception of the initial preparation. a lawyer came to me and asked. This is why any subsequent measurement of σz will provide the same outcome: One the state vector is collapsed to |0 . This is different in case we measure σx in between.3. yet. well. for the same reasons – only with the roles of the two vectors reversed. This notion of a measurement of σx “disturbing” a measurement of σz is a manifestation of the observables σx and σz not commuting: They are incompatible. the wave function is projected onto |0 or |1 . (2. the probability of obtaining the outcome 1 is again p0 = 1. Recently.99) A fancy way of saying this is that σz and σz are compatible. actually. Such properties are simply not defined within quantum theory. and will get the same value over and over again. we have obtained the outcome λ0 = 1 and the post-measurement state |0 . no. Compatible observables can be repeatedly measured without altering the outcome. if we obtain |+ in the measurement of the observable σx . QUANTUM STATES AND OBSERVABLES 2. that “nothing is precisely defined”? Well. the above situation is very clear. while this is not so for incompatible observables. and it does simply not make sense to ask whether the spin is “truly pointing up or down” in case one has just performed a measurement along the x axis. .4 Three readings of the Heisenberg uncertainty relation This notion of outcomes of measurements being “uncertain” can be made more precise in terms of the so-called uncertainty relation. λ1 = ±1. then the probability of getting |0 in a subsequent σz measurement is again 1/2. The first measurement of the Pauli operator σz can have the outcomes λ0 . After the measurement. Then. is the uncertainty in adjudication not just a manifestation of the Heisenberg principle. dependent on the measurement outcome. The order of measurement also matters.100) 2. that are so often misunderstood as the uncertainty relation. Say.3. say. | −|0 |2 = 1 .

The central object here is the mean square deviation of an observable from its expectation value. ψ |(CD)† |ψ = z ∗ = x − iy.107) as numbers always commute with each other. (2. We now make use of the CauchySchwarz inequality – a useful inequality from linear algebra. Taking the absolute value.102) We now consider this for two observables A and B .108) (2.2. (2. and the expectation value can be complex. we obtain | [C. MEASUREMENT 21 But except from the principle being overly referred to in urban slang. B− B . we have for the commutator [C. y ∈ R. Later I understood that this is because there have been different readings used of the principle.105) (2. Of course. D] = ψ |[C. (2. even within the formalism there often is a kind of confusion coming along with this principle. Let us write CD DC = = ψ |CD|ψ = z = x + iy. and slightly irritated by the fact that nobody seemed to have any desire to fix the precise prefactor on the right hand side of the inequality. (2. D] |2 ≤ 4| ψ |CD|ψ |2 . (∆A)2 = (A − A )2 = A2 − A 2 . let us define C D = = A− A .104) Even if C and D are clearly Hermitian. That is to say. for a system initially prepared in |ψ .110) We can now collect the terms and state the uncertainty relation: . We will mention three readings of Heisenberg’s uncertainty principle.103) (2. B ]. This leads to | ψ |CD|ψ |2 = |( ψ |C ) · (D|ψ )|2 ≤ = C |ψ 2 2 · D|ψ 2 2 ψ |C |ψ · ψ |D |ψ . even if we discuss in detail only one.109) (2. [C. D]|ψ = z − z ∗ = 2iy = 2i im( CD ). For simplicity of notation. the product CD is usually no longer Hermitian. I was slightly confused when I first heard of it in my first lecture. (2. D] = [A.106) with x.3.

This means that if we prepare many quantum systems and first estimate ∆A many times. One can also derive uncertainties for the Heisenberg uncertainty principle in this second reading. the explanation given is quite often incompatible with the above derivation. the product of the two uncertainties of A and B cannot be arbitrarily small.) of the initial preparation). the product of the two uncertainties cannot take an arbitrarily small value – unless the two observables A and B commute. Quantum mechanics simply does not allow for any smaller uncertainties. Interestingly. This is the notion of a measurement of one observable A makes the outcome of another non-commuting observable B less certain. So in a way. When applied to position and momentum. and then estimate ∆B : Then the product of the two uncertainties will always be larger than the above given value. 2 (2. B ]|ψ |. This is the second reading of the uncertainty relation. If we “know the value of A precisely.i.d. QUANTUM STATES AND OBSERVABLES Heisenberg’s uncertainty relation: Two observables A and B of a quantum system prepared in |ψ satisfy the uncertainty relation ∆A · ∆B ≥ 1 | ψ |[A. but this is not quite what we have derived above (which made use of the independent and identical distribution (i. such that we know it later with high statistical significance. This is the reading of the famous Heisenberg microscope: Heisenberg discussed how a measurement of X disturbs later later measurements of the non-commuting observable P . however. Precisely. . then B . This is a remarkable observation. is that one independently prepares the systems and measures either A or B .112) which means that wave functions cannot be arbitrarily narrow both in position and momentum. one cannot “know the values of two non-commuting observables at once”. This is the first reading of the Heisenberg uncertainty relation.111) That is to say. then the measurement of B will be a lot disturbed”. A third reading which we only briefly mention is the one where tries to jointly measure two observables in a single generalized measurement. what is meant. This is also the derivation that most books on quantum mechanics offer.22 CHAPTER 2. but the prefactor on the right hand side will be slightly different. Otherwise. This is also true. Just for the records: Here one aims at obtaining as much information as possible about A and B in a single run of a measurement. this principle reads ∆X · ∆P ≥ 2 . (2. We have to delay this discussion as we are at this point not quite clear about what a generalized measurement is. We hence measure on the same system first A. and then analyses the data.

MEASUREMENT 23 which is then repeated many times. This again leads to an uncertainty relation. All these readings have the narrative in common.3. however.2. that if the mean square deviation of one observable is small. . but yet again with a different prefactor. it cannot be small at the same time for a non-commuting observable. This is again one of the profound consequences of non-commuting operators in quantum theory.